|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* Plugin: SEOToolbox |
|
4
|
|
|
* Author: Dylan Grech |
|
5
|
|
|
* Copyright: 2016 |
|
6
|
|
|
* |
|
7
|
|
|
* SEOtoolbox Controller Extension decorates the Content Controller |
|
8
|
|
|
* to add the auotamted links to a page where needed |
|
9
|
|
|
* |
|
10
|
|
|
* @see AutomatedLink |
|
11
|
|
|
*/ |
|
12
|
|
|
class SEOToolboxControllerExtension extends Extension { |
|
13
|
|
|
|
|
14
|
|
|
private $maxLinksPerPage; |
|
15
|
|
|
private $settings = null; |
|
16
|
|
|
private $linkCount = 0; |
|
17
|
|
|
private $addLinks = true; |
|
18
|
|
|
private $excludeTags = array(); |
|
19
|
|
|
private $maxLinks = 0; |
|
20
|
|
|
|
|
21
|
|
|
public function index(){ |
|
22
|
|
|
$this->addAutomatedLinks(); |
|
23
|
|
|
|
|
24
|
|
|
// If we have a crawl request check the CrawlID so we're sure we didn't hit another SS site running our module |
|
25
|
|
|
if( $crawl_id = $this->owner->request->getHeader('X-Crawl-Id') ){ |
|
26
|
|
|
return( $crawl_id == GlobalAutoLinkSettings::get_current()->CrawlID ) |
|
27
|
|
|
? $this->crawl_response() |
|
28
|
|
|
: $this->owner->redirect(SEOTestSiteTreeController::getPermissionDeniedPage()->Link()); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
return array(); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
private function crawl_response(){ |
|
35
|
|
|
// Encoded version to detect which fields are being used |
|
36
|
|
|
$customize = array(); |
|
37
|
|
|
$dbFields = Config::inst()->get($this->owner->ClassName, 'db'); |
|
38
|
|
|
if(is_array($dbFields)) { |
|
39
|
|
|
foreach ( $dbFields as $field => $type) { |
|
40
|
|
|
if (strtolower($type) == 'htmltext') { |
|
41
|
|
|
$data = ($this->owner->hasMethod($field)) ? $this->owner->$field() : $this->owner->$field; |
|
42
|
|
|
if($data){ |
|
43
|
|
|
$tmp = new HTMLText('tmp'); |
|
44
|
|
|
$tmp->setValue($data); |
|
45
|
|
|
$data = base64_encode($tmp->forTemplate()); |
|
46
|
|
|
$customize[$field] = "[**[$field]**[$data]**]"; |
|
47
|
|
|
} |
|
48
|
|
|
} |
|
49
|
|
|
} |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
if (in_array($this->owner->ClassName, ClassInfo::subclassesFor('ErrorPage'))) { |
|
53
|
|
|
header("HTTP/1.0 405 Instance of ErrorPage"); |
|
54
|
|
|
die(); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
// Clean out the html before sending it back to minimize response size |
|
58
|
|
|
die( |
|
59
|
|
|
preg_replace(array( |
|
60
|
|
|
'/<style(.*?)[>]/im', |
|
61
|
|
|
'/<(script|noscript)(.*?)<\/(script|noscript)[>]/im', |
|
62
|
|
|
'/<!--(.*?)-->/im', |
|
63
|
|
|
), '', $this->owner->customise($customize)->render()) |
|
64
|
|
|
); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Get the global settings and check if we should be adding |
|
69
|
|
|
* links to this page |
|
70
|
|
|
* |
|
71
|
|
|
* @return GlobalAutoLinkSettings|false |
|
72
|
|
|
*/ |
|
73
|
|
|
private function getSettings() { |
|
74
|
|
|
if ($this->settings === null) { |
|
75
|
|
|
$this->settings = GlobalAutoLinkSettings::get_current(); |
|
76
|
|
|
if (!$this->settings) return $this->addLinks = false; |
|
77
|
|
|
|
|
78
|
|
|
$this->excludeTags = (array) $this->settings->ExcludeTags(); |
|
79
|
|
|
$this->maxLinks = (int) ($this->settings->MaxLinksPerPage) ? $this->settings->MaxLinksPerPage : PHP_INT_MAX; |
|
80
|
|
|
|
|
81
|
|
|
if (!in_array($this->owner->ClassName, $this->settings->AllowedIn())) $this->addLinks = false; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
return $this->settings; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
/** |
|
88
|
|
|
* Goes through all the automated link settings and adds |
|
89
|
|
|
* the links where necessary |
|
90
|
|
|
* |
|
91
|
|
|
* @return void |
|
92
|
|
|
*/ |
|
93
|
|
|
public function addAutomatedLinks(){ |
|
94
|
|
|
if( GlobalAutoLinkSettings::$enabled && $this->owner->class != 'RedirectorPage' ) { |
|
95
|
|
|
$this->getSettings(); |
|
96
|
|
|
if( !$this->addLinks ) { |
|
97
|
|
|
return; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
foreach( $this->getSettings()->IncludeInFields() as $field ){ |
|
101
|
|
|
// Check that the field provided by user exists in this object, is of type HTMLText and has content |
|
102
|
|
|
if( AutomatedLink::isFieldParsable( $this->owner->data(), $field ) ){ |
|
103
|
|
|
|
|
104
|
|
|
// Create dummy object so we can parse the HTML |
|
105
|
|
|
$dummy = new HTMLText( $field ); |
|
106
|
|
|
$dummy->setValue( $this->owner->$field ); |
|
107
|
|
|
// Create DOMDocument Object |
|
108
|
|
|
$content = mb_convert_encoding( $dummy->forTemplate(), 'html-entities', GlobalAutoLinkSettings::$encoding ); |
|
109
|
|
|
$dom = AutomatedLink::constructDOMDocument($content); |
|
110
|
|
|
|
|
111
|
|
|
// Check current link count and if it's already exceeded do nothing |
|
112
|
|
|
$this->linkCount += (int) $dom->getElementsByTagName( 'a' )->length; |
|
113
|
|
|
if( $this->linkCount >= $this->maxLinks ) { |
|
114
|
|
|
return; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
$parsed = $this->parseField( $dom, $field ); |
|
118
|
|
|
$this->owner->data()->$field = $parsed; |
|
119
|
|
|
$this->owner->$field = $parsed; |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* Goes through $tags and replaces them in $html with a hash of themselves. |
|
127
|
|
|
* Returns an array of hash keys used and the original value |
|
128
|
|
|
* |
|
129
|
|
|
* @param DOMDocument $html |
|
130
|
|
|
* @param array $hash_tags |
|
131
|
|
|
* @return array |
|
132
|
|
|
*/ |
|
133
|
|
|
private function hashTags( DOMDocument $html, $hash_tags ){ |
|
134
|
|
|
$excluded = array(); |
|
135
|
|
|
foreach( $hash_tags as $eTag ){ |
|
136
|
|
|
while( $tags = $html->getElementsByTagName( $eTag ) ){ |
|
137
|
|
|
if( !$tags->length ) break 1; |
|
138
|
|
|
$tag = $tags->item(0); |
|
139
|
|
|
$value = $html->saveHTML( $tag ); |
|
140
|
|
|
$key = (string) crc32( $value ); |
|
141
|
|
|
|
|
142
|
|
|
// Convert back children nodes of this node if they were already hashed |
|
143
|
|
|
$excluded[$key] = str_replace( array_keys( $excluded ), array_values( $excluded ), $value ); |
|
144
|
|
|
|
|
145
|
|
|
$tag->parentNode->replaceChild( $html->createTextNode( $key ), $tag ); |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
return $excluded; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
/** |
|
153
|
|
|
* Adds the passed automated link to $content if possible. |
|
154
|
|
|
* Returns an array of hashed that have been added and the original value that was replaced |
|
155
|
|
|
* |
|
156
|
|
|
* @param AutomatedLink $link |
|
157
|
|
|
* @param string $content |
|
158
|
|
|
* @return array |
|
159
|
|
|
*/ |
|
160
|
|
|
public function addLinkToContent(AutomatedLink $link, &$content){ |
|
161
|
|
|
$links = array(); |
|
162
|
|
|
$max = (int) ( $link->MaxLinksPerPage > 0 ) ? $link->MaxLinksPerPage : PHP_INT_MAX; |
|
163
|
|
|
$escape = (string) preg_quote( $link->Phrase, '/' ); |
|
164
|
|
|
$regex = (string) ( $link->CaseSensitive ) ? "/(\b{$escape}\b)/" : "/(\b{$escape}\b)/i"; |
|
165
|
|
|
|
|
166
|
|
|
// Count the matches |
|
167
|
|
|
preg_match_all( $regex, $content, $count ); |
|
168
|
|
|
$count = ( is_array( $count ) && isset( $count[0] ) ) ? count( $count[0] ) : 0; |
|
169
|
|
|
if( $count < 1 ) $links; |
|
170
|
|
|
|
|
171
|
|
|
if( isset( $this->maxLinksPerPage[ $link->ID ] ) ) { |
|
172
|
|
|
$max -= $this->maxLinksPerPage[$link->ID]; |
|
173
|
|
|
} else { |
|
174
|
|
|
$this->maxLinksPerPage[$link->ID] = 0; |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
for( $x = 0; $x < $count; $x++ ){ |
|
178
|
|
|
// Stop adding links if we reached the link or page limit |
|
179
|
|
|
if( $x >= $max || $this->linkCount >= $this->maxLinks ) break; |
|
180
|
|
|
|
|
181
|
|
|
// Check if there is anything else to replace else stop |
|
182
|
|
|
preg_match( $regex, $content, $match ); |
|
183
|
|
|
if( !is_array( $match ) || !count( $match ) ) break; |
|
184
|
|
|
|
|
185
|
|
|
if( !$html = (string) $link->getHTML( $match[0] ) ) continue; |
|
186
|
|
|
$key = (string) crc32( $html ); |
|
187
|
|
|
$links[ $key ] = (string) $html; |
|
188
|
|
|
|
|
189
|
|
|
$content = preg_replace( $regex, $key, $content, 1 ); |
|
190
|
|
|
$this->linkCount++; |
|
191
|
|
|
$this->maxLinksPerPage[ $link->ID ]++; |
|
192
|
|
|
} |
|
193
|
|
|
|
|
194
|
|
|
return $links; |
|
195
|
|
|
} |
|
196
|
|
|
|
|
197
|
|
|
/** |
|
198
|
|
|
* Parse the provided field and add the necessary links |
|
199
|
|
|
* |
|
200
|
|
|
* @param DOMDocument $html |
|
201
|
|
|
* @param String $field |
|
202
|
|
|
* @return string |
|
203
|
|
|
*/ |
|
204
|
|
|
private function parseField( DOMDocument $html, $field ){ |
|
205
|
|
|
$this->owner->extend( 'beforeParseField', $html, $field ); |
|
206
|
|
|
|
|
207
|
|
|
// Remove Tags from Content we wown't be using |
|
208
|
|
|
$excluded = $this->hashTags($html, $this->excludeTags); |
|
209
|
|
|
$body = (string)$html->saveHTML( $html->getElementsByTagName('body')->item(0) ); |
|
210
|
|
|
$content = preg_replace( array( '/\<body\>/is', '/\<\/body\>/is' ), '', $body, 1 ); |
|
211
|
|
|
|
|
212
|
|
|
foreach( AutomatedLink::get()->sort('Priority') as $link){ |
|
213
|
|
|
if( $this->linkCount < $this->maxLinks && $link->canBeAdded( $this->owner, $field ) ) { |
|
214
|
|
|
$links = $this->addLinkToContent($link, $content); |
|
215
|
|
|
if( is_array($links) && count($links) > 0 ) { |
|
216
|
|
|
$excluded = $excluded + $links; |
|
217
|
|
|
} |
|
218
|
|
|
} |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
// Re-add the excluded Tags |
|
222
|
|
|
return str_replace( array_keys( $excluded ), array_values( $excluded ), $content ); |
|
223
|
|
|
} |
|
224
|
|
|
} |
|
225
|
|
|
|
dylangrech92 /
seotoolbox
